This is an experiment!
The deepseek v4 flash model and the qwen3.6 27b model were merged.
This is a cross-architecture model merge, with the architecture aligned to the QWEN 3.5 architecture.
The model is indeed usable.
The deepseek v4 flash model and the qwen3.6 27b model were merged.
This is a cross-architecture model merge, with the architecture aligned to the QWEN 3.5 architecture.
The model is indeed usable.